# Type a three-word title into three corners, one word per beat, over a wide shot

Manual DaVinci Resolve / Fusion construction guide generated from the recipe. It is not a tested .drp.
The accompanying Remotion project (recipe.json) is the executable definition; the numbers below are copied from it.
Neither construction establishes source fidelity: this is an original vector approximation of an inspected editing moment.

## Source and clock
Reference: OpenAI's "Building with ChatGPT Voice | OpenAI" (E0ZMOschrTU), published 2026-07-23.
Source interval [122, 175) on the 24000/1001 fps master: 53 frames, 2.210542 s. Local frame 0 = source frame 122.
Create a 1920x1080 progressive timeline at 24000/1001 fps and a Fusion composition exactly 53 frames long. Keep integer frames; do not round the frame rate.
Background field: #7a6a5b. Export silent H.264 yuv420p to match the intended playback container.

## Observation
Hard cut to a high, wide shot of a living room. 'Building' types on character by character at the top-left (f2-10), then 'with' types on at the right edge mid-frame (f14-22), then 'voice' types on at the bottom-right (f22-30). All three words hold in place with a slight camera drift until a hard cut at f53.

## Mechanism
A staggered three-word title: each word types on in its own corner in sequence, building a triangular composition over live footage.

## Phases (local frames)
- 0: Hard cut to wide room shot
- 2: Word one types on, top-left
- 14: Word two types on, right-middle
- 22: Word three types on, bottom-right
- 31: Hold until the cut

## Layer inventory (absolute 1920x1080 top-left pixels, back to front)
- room [group] at (0, 0) size 1920x1080
    keyframes scale: f0=1.0, f53=1.02
  - floor [rect] at (0, 0) size 1920x1080 fill #7a6a5b
  - rug [rect] at (120, 300) size 1500x620 radius 8 fill #8e8074
  - sofa [rect] at (160, 330) size 900x180 radius 26 fill #a15a2e
  - chair [ellipse] at (1120, 300) size 220x220 fill #b8582a
  - table [rect] at (640, 600) size 700x260 radius 10 fill #b58a4e
  - laptop [rect] at (1120, 640) size 150x100 radius 6 fill #2b2b2b
  - plant [ellipse] at (1560, 380) size 180x200 fill #5c7a4a
  - wall [rect] at (1400, 0) size 520x160 fill #8f8a80
  - screen [rect] at (1020, 40) size 260x140 radius 6 fill #e7cfd0
- w1 [text] at (70, 56) text "Building" 92px weight 600 #ffffff reveal chars frames 2-10
- w2 [text] at (1200, 500) size 650x0 text "with" 92px weight 600 #ffffff reveal chars frames 14-22
- w3 [text] at (1200, 880) size 650x0 text "voice" 92px weight 600 #ffffff reveal chars frames 22-30

## Construction steps
1. Timeline 1920x1080 at 23.976 fps; Fusion composition 53 frames. Local 0 = master frame 122 (the cut into the wide shot).
2. Footage layer: the wide room plate; add a Transform with Size 1.00 to 1.02 across the whole clip for the slow drift.
3. Three Text+ nodes, 92px, weight 600, white: word one left-aligned at (70,56); word two right-aligned to x=1850 at y=500; word three right-aligned to x=1850 at y=880.
4. Write On keyframes (chars): word one 0 to 1 over frames 2-10, word two over 14-22, word three over 22-30; no fade, no caret.
5. Easy mistake: word two and three start on the same frame word one would have finished; keep the 4-frame gap after word one and let word three begin exactly as word two completes so the stagger reads as a rhythm, not three separate cues.
6. Hold all three words to the last frame; the source ends on a hard cut, not a fade.

## Content variants
Default copy:
- w1: Building
- w2: with
- w3: voice
Adaptation changes:
- w1: Shipping
- w2: in
- w3: public

## Limits
Easing names map to Fusion spline shapes: linear = linear, in/out/inOut = ease curves on the same frames, hold = step, spring/overshoot = slight overshoot past the target before settling.
Text reveal frames give the first and last unit; distribute characters/words evenly between them. Cursor click rings last 14 frames.
Vector illustration is an approximation of layout and motion, not the product UI or recovered production assets.
